Silvia Salini

Bayesian networks of customer satisfaction survey data

A Bayesian network (BN) is a probabilistic graphical model that represents a set of variables and their probabilistic dependencies. Formally, BNs are directed acyclic graphs whose nodes represent variables, and whose arcs encode the conditional dependencies among the variables. Nodes can represent any kind of variable, be it a measured parameter, a latent variable, or a hypothesis. They are not restricted to represent random variables, which form the “Bayesian” aspect of a BN. Efficient algorithms exist that perform inference and learning in BNs. BNs that model sequences of variables are called dynamic BNs. In this context, [A. Harel, R. Kenett, and F. Ruggeri, Modeling web usability diagnostics on the basis of usage statistics, in Statistical Methods in eCommerce Research, W. Jank and G. Shmueli, eds., Wiley, 2008] provide a comparison between Markov Chains and BNs in the analysis of web usability from e-commerce data. A comparison of regression models, structural equation models, and BNs is presented in Anderson et al. [R.D. Anderson, R.D. Mackoy, V.B. Thompson, and G. Harrell, A bayesian network estimation of the service–profit Chain for transport service satisfaction, Decision Sciences 35(4), (2004), pp. 665–689]. In this article we apply BNs to the analysis of customer satisfaction surveys and demonstrate the potential of the approach. In particular, BNs offer advantages in implementing models of cause and effect over other statistical techniques designed primarily for testing hypotheses. Other advantages include the ability to conduct probabilistic inference for prediction and diagnostic purposes with an output that can be intuitively understood by managers.

Robust analysis of bibliometric data

This work stems from the idea of describing the scientific productivity of Italian statisticians. There are several problems that must be addressed in achieving this goal: What data should be used? Have the data been cleaned? What techniques can be used? We propose the use of multiple sources and multiple metrics to get a complete information base. We check the correctness of the data using multivariate outlier identification techniques. We appropriately transform the data. We apply robust clustering to verify the existence of homogeneous groups. We suggest the use of forward search to establish a ranking among scholars. The proposed methodology, which, in this case, allowed us to group scholars into four homogeneous groups and sort them according to multidimensional data, can be applied to other similar applications in bibliometrics.

The Rasch Model to Measure Service Quality

In this paper the Rasch model will be considered, a statistical tool arising from psychometric field, which allows the examination of the service quality through the known variables: for each of those an objective measure is obtained. In particular, the application of Rasch model will concern the quality of university teaching. Moreover a method will be proposed to obtain a segmentation of student population only based on satisfaction index

The Rasch Model in Customer Satisfaction Survey Data

Abstract This paper deals with the measurement of a service or product quality using Customer Satisfaction Survey results. Many different methods are used to analyse customer satisfaction data. Some use statistical models which estimate the relationship between the latent and manifest variables (LISREL, PLS, etc.), whilst others use dimensionality reduction methods (FA, PCA, etc.). All of these methods require a numerical quantification of the categories and consequently the distance between the numerical labels is fixed and the linear relationship between the variables implicitly assumed. Moreover these methods produce a customer satisfaction measure for each subject and an evaluation of its importance on the satisfaction level for each item. When analyzing quality and satisfaction levels together, the Rasch model (RM) appears to be particularly appropriate. A Likert scale is not required and non-linear relationships are involved. Moreover, a Rasch analysis can also act as a useful diagnostic tool for calibrating the questionnaire itself. In this paper we will present three different applications of the Rasch Model for the purposes of measuring quality and customer satisfaction levels. For each technique we will highlight its peculiarities, give an interpretation of the parameters used, analyse the model’s fit with the data and perform a critical analysis of the results.

Customer satisfaction in the airline industry : the case of British Airways

In this paper two different non-classic methods, based on the analysis of qualitative data, are applied to evaluate customer satisfaction. The airline industry is considered with British Airways used as a case study. First, a classification algorithm based on the decision tree theory is performed. By conserving the original ordinal measuring scale for the items, it is possible to rank the dimensions in order to obtain a map of preference, select the groups of subjects more or less satisfied and delineate the peculiar characteristics of each group. Second, a probabilistic model, the Rasch model, is applied with the aim of distinguishing the two components (satisfaction and quality) that influenced the answer to each item and obtaining a numerical interval measure for both components. Copyright

Ten challenges in modeling bibliographic data for bibliometric analysis

The complexity and variety of bibliographic data is growing, and efforts to define new methodologies and techniques for bibliometric analysis are intensifying. In this complex scenario, one of the most crucial issues is the quality of data and the capability of bibliometric analysis to cope with multiple data dimensions. Although the problem of enforcing a multidimensional approach to the analysis and management of bibliographic data is not new, a reference design pattern and a specific conceptual model for multidimensional analysis of bibliographic data are still missing. In this paper, we discuss ten of the most relevant challenges for bibliometric analysis when dealing with multidimensional data, and we propose a reference data model that, according to different goals, can help analysis designers and bibliographic experts in working with large collections of bibliographic data.

Optimal Scaling for Risk Assessment: Merging of Operational and Financial Data

The focus of the paper is the use of optimal scaling techniques to reduce the dimensionality of ordinal variables describing the quality of services to a continuous score interpretable as a measure of operational risk. This new score of operational risk is merged with a financial risk score in order to obtain an integrated measure of risk. The proposed integration methodology is a generalization of the merging model suggested in Fagini and Giudici (J. Oper. Res. Soc. 2010; in press) for a hierarchical data structure. In order to demonstrate the methodology, we use real data from a telecommunication company providing services to enterprises in different business lines and geographical locations. For each enterprise, we have collected information about operational and financial performance. The approach demonstrated in this case study can be generalized to general service providers who are concerned by both the quality of service and the financial solvency of their customers. Copyright

Migraine with aura with onset in childhood and adolescence: Long-term natural history and prognostic factors

The long-term course of migraine with aura (MA) has been poorly explored. The present 11-year follow-up study assessed the long-term natural history and possible prognostic factors of MA with onset in childhood or adolescence. Patients were recruited from the original case records of our department, which are specifically designed to report all headache characteristics, aura symptoms and electroencephalogram (EEG) findings. A total of 77 patients (47 females; 30 males) whose records contained detailed descriptions of both headache and aura symptoms apparently meeting the International Classification of Headache Disorders (ICHD)-II criteria for MA (i.e., 1.2.1, 1.2.2, 1.2.6) underwent structured face-to-face follow-up headache interviews, all of which were conducted by the same neurologist, who has particular expertise in this field. A multivariate model (logistic regression analysis) was used to investigate the association between possible prognostic factors and the remission of both aura and headache at follow-up. The results of our study showed that 23.4% of the MA patients were headache-free at follow-up, 44.1% still had MA and 32.5% had a transformed headache diagnosis (i.e., fulfilling the criteria for ICHD-II 1.1. or 2). Patients with basilar-type migraine (1.2.6) showed the highest headache remission rate (38.5%). Our study seems to show that migraine with typical aura (1.2.1–1.2.2) is associated with a favourable evolution of aura symptoms over time (remission of aura in 54.1% of patients). Subjects experiencing only visual aura had a lower remission rate compared with those with visual ± sensory ± aphasic aura symptoms (36.8% vs. 61.5%, p = 0.054). A short headache duration (<12 hrs) and the presence of EEG abnormalities at baseline were the only significant predictors of aura remission at follow-up (odds ratio [OR] = 9.12, 95% confidence interval [CI]: 1.79 ± 46.51, and OR = 4.76, 95% CI: 1.18 ± 19.15, respectively). No significant predictors of headache remission were found. In conclusion, our results suggest that MA shows a favourable course. Further prospective studies with detailed EEG analysis both at baseline and at follow-up are needed in order to confirm the possible prognostic role of EEG abnormalities in MA. That said, it would, in our opinion, be highly premature at present to submit children with MA to EEG examinations for prognostication purposes.

The Rorschach test and Gilles de la Tourette’s syndrome: A pilot case–control study

BACKGROUND To date only three studies, all lacking a control group, have investigated the personality style underlying tic disorders or Tourettes syndrome (TS), using the projective Rorschach test. Despite the recent controversy about its proper use in clinical practice and research, the Rorschach test may be useful to explore personality organization in TS. The research hypothesis under investigation in this exploratory study was that young TS patients exhibit a peculiar personality organization, with significantly increased indicators of poor emotional control, aggression, anxiety and depression, on the Rorschach test, compared to healthy controls. METHODS We analyed the Rorschach protocols of 17 newly diagnosed paediatric TS patients and 17 age- and sex-matched healthy controls. RESULTS Compared with the controls young TS patients recorded significantly lower mean percentages of F and F+ responses and a significantly higher number of FM and FC responses, thus showing reduced control of emotions and difficulty in integrating aggressivity. Furthermore, associations emerged between comorbidity with ADHD and increased emotional instability and between severity of tics and of OCD and a pattern of emotional hyper-control and application of rigid defensive mechanisms. CONCLUSIONS These results seem to confirm the existence of a peculiar personality style in young TS patients and may have important implications, both pathogenetic and therapeutic.

A decade of research in statistics: a topic model approach

Topic models are a well known clustering approach for textual data, which provides promising applications in the bibliometric context for the purpose of discovering scientific topics and trends in a corpus of scientific publications. However, topic models per se provide poorly descriptive metadata featuring the discovered clusters of publications and they are not related to the other important metadata usually available with publications, such as authors affiliation, publication venue, and publication year. In this paper, we propose a methodological approach to topic modeling and post-processing of topic models results to the end of describing in depth a field of research over time. In particular, we work on a selection of publications from the international statistical literature, we propose an approach that allows us to identify sophisticated topic descriptors, and we analyze the links between topics and their temporal evolution.